An Improved Multi-scale Retinex Fog and Haze Image Enhancement Method

Haze is a common phenomenon in major cities in recent years, image taken in foggy weather are blurred inview, which seriously impact the application of traffic system and outdoor visions system. The haze image frequency, intermediate frequency, low frequency and spectrum image are analyzed separately first in the paper, and then by fixed Retinex light reflection model, a new haze image enhancement method is proposed based on the multi-scale Retinex algorithm MSR. Improved Retinex algorithm is proposed to enhance image contrast of this method. The irradiated component and the reflected component of the image of non-liner transformation and then linear contrast stretch for the output image. An experimental result has show that the image contrast can be enhance effectively and achieve haze image enhancement through this method.